Aquaculture 467 (2017) 49–62

Contents lists available at ScienceDirect

Aquaculture

j ourna l homepage: www.e lsev ie r .com/ locate /aquacu l tu re

Future availability of raw materials for salmon feeds and supply chainimplications: The case of Scottish farmed salmon

C. Jonathan Shepherd a, Oscar Monroig b,⁎, Douglas R. Tocher b

a Bluetail Consulting Ltd., Richmond, Surrey TW9 3NL, UKb Institute of Aquaculture, Faculty of Natural Sciences, University of Stirling, Stirling FK9 4LA, UK

⁎ Corresponding author.E-mail address: oscar.monroig@stir.ac.uk (O. Monroig

http://dx.doi.org/10.1016/j.aquaculture.2016.08.0210044-8486/© 2016 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 15 December 2015Received in revised form 10 August 2016Accepted 16 August 2016Available online 17 August 2016

The current range of Scottish salmon feeds is adapted to a differentiated supply of salmonproducts, including dif-fering omega-3 content, differing content of marine raw materials, etc. The progressive replacement of marinefeed ingredients by plant proteins and oils is reducing the content of omega-3 long-chain polyunsaturatedfatty acids (LC-PUFA). However the benefits are a more secure and less volatile raw material supply, togetherwith environmental feed contaminants at low or undetectable levels in the resulting salmon product. There iswidespread adoption of standards and certification schemes by Scottish salmon farmers and feed suppliers inorder to demonstrate environmental sustainability. This has focused in particular on use of certified ingredientsfrom sustainable supply sources (‘responsible sourcing’). Future volume estimates of Scottish salmon production,hence feed requirements, are insufficient to threaten raw material supply compared with global markets, al-though it is argued this is likely to involve greater use of locally grown plant proteins and an increased proportionof fishmeal manufactured from by-product trimmings (derived from processing fish for human consumption).However, UK retail chains will remain reluctant to allow salmon suppliers to utilise land animal by-productsdue to negative consumer perceptions, with resulting implications for formulation cost and flexibility. Given itsworld-wide scarcity, the main strategic concern relates to future availability of sufficient omega-3 LC-PUFA, inparticular eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), in order to maintain the healthyimage of Scottish salmon. To maintain its longer-term reputation and product benefits, the Scottish industrymay need to consider adopting a more flexible attitude to using new alternatives to fish oil (e.g. EPA and DHAderived from transgenic oil seed crops, when commercially available). It is concluded that Scottish salmon farm-ing is a successful example of sustainable feed development and the industry can be confident that the changingrawmaterial base will support continuing production of high quality, healthy farmed salmon, but the long-termsecurity of supply of omega-3 LC-PUFA remains an issue.

© 2016 Elsevier B.V. All rights reserved.

Keywords:Atlantic salmonSustainable feedsFishmealFish oilAlternative proteinsAlternative oilsOmega-3EPADHASupply chains

1. Introduction to Scottish salmon feeds

1.1. Study aims and methodology

Farmed Scottish salmon has a distinct market position, which hasimplications for feed requirements. The purpose of this study was toconsider trends in the change in composition of Scottish salmon feedswith regard to the availability and use of key feed ingredients, and toconsider the resulting implications for farmed salmon supply chains.In particular the study aim was to identify the nutritional, ethical, mar-keting, and sustainability issues surrounding reduced reliance of thesalmon farming industry on marine ingredients. In order to achievethis, over the period from April 2014 to March 2015, meetings and

).

telephone interviews were conducted with industry experts and stake-holders in the UK, Norway, Denmark, France, Ireland, and the USA.

1.2. Global farmed salmon industry and the differentiated supply of Scottishsalmon and feeds

Salmon farming is themost highly developed form of large-scale in-tensive aquaculture owing to its productivity growth and technologicalchange since the industry started in the 1970′s in Norway and Scotland;it represents a highly efficient mechanism for transforming marine re-sources into high quality food that is available on a year round basis(Hognes et al., 2011; Ytrestøyl et al., 2011). Estimated global productionof farmed Atlantic salmon in 2013was approx. 1.8million tonnes and isdominated by Norway with 1.1 million tonnes (Kontali, 2013). Themuch smaller Scottish industry (163,234 tonnes in 2013, Table 1) close-ly parallels Norway in terms of farming technology and is nowmainly inNorwegian ownership. The compound annual growth rate (CAGR) for

Table 1Comparisonof farmed salmonproduction (tonnes), feed supply volume (tonnes), and cor-responding feed conversion ratio (FCR) for Scotland and Norway in years 2013 and 2014.Source: Kontali AS.

2013 2014

Scotland Norway Scotland Norway

Farmed salmon production as wholefish equivalent

163,234 1,143,500 179,022 1,198,900

Salmon feed supply 214,000 1,487,600 221,000 1,598,800Feed conversion ratio 1.31 1.30 1.23 1.33

65.4

33.524.8 19.5 18.3

24.0

31.1

16.6

11.2 10.9

22.2

35.5

36.7 36.7

12.518.3 19.2

9.6 11.2 8.4 11.1 11.2

1.0 2.0 2.2 3.1 3.7

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

1990 2000 2010 2012 2013

Ingredient sources (%) 1990-2013

microingredients

starch

plant oil

plant protein

marine oil

marine protein

50 C.J. Shepherd et al. / Aquaculture 467 (2017) 49–62

farmed Atlantic salmonwas 6% on a global basis from 2003 to 2013, butis nowdecreasing (MarineHarvest, 2014). In both Scotland andNorwaythe growth in salmon farming has resulted in year-round availability ofcost-efficient, affordable salmon products on a greatly increased scale,while creatingwealth and sustainable employment, including to remoterural locations; the industry therefore demonstrates clear economic andsocial benefits (Asche and Bjørndal, 2011).

Table 1 compares Scottish with Norwegian farmed salmon produc-tion, feed supply, and feed conversion rate (FCR) for 2013 and 2014.At c. 13% of combined Scottish and Norwegian production volume in2014, it is clear that Scottish production is relatively small. Also the Scot-tish Government's target for sustainable growth of marine finfish is210,000 tonnes by 2020. It should be noted that 2014 Scottish salmonproductionwas boosted by premature 4th quarter harvesting for healthreasons; as a consequence an FCRof 1.23 appears unrealistically low andindustry sources indicate it was close to the 2013 value at c. 1.3. Feedcosts dominate production costs for farmed salmon with the 2013 costof salmon feed in Norway and Scotland for Marine Harvest (the largestglobal salmon producer) estimated at 50% and 46% respectively of pro-duction costs; the corresponding figures of 41% and 43% respectively forCanada andChile aremainly due to lower feed costs linked to use of landanimal by-products (Marine Harvest, 2014).

Despite being relatively small in volume comparedwithNorway andChile, the Scottish salmon industry holds a distinctive position in worldmarkets. Whereas the Norwegian salmon farming industry is the lead-ing world producer and exporter of farmed salmon as a standardisedcommodity product, including to the UK market, the Scottish industry,lacking comparable economies of scale, supplies more differentiatedproducts at higher unit value to offset potentially higher productioncosts. These range from standard to high performance products1, in-cluding bespoke supply for niche markets in the UK and overseas(mainly to the USA and France). Scottish production therefore relieson a variety of different feed products, including specialist formulationsto support a premium salmon segment, i.e. higher price salmon prod-uctswith quality characteristics, including label claims on Scottish prov-enance, the content of omega-3 LC-PUFA in regard to EU intakerecommendations, and others (e.g. compliance with responsible farm-ing standards).

2. Changing salmon feed composition

2.1. Historical picture and trends

Historically the twomost important ingredients in salmon feed havebeen fishmeal and fish oil, having favourable nutrient compositions andreflecting a major food item for a carnivorous species like salmon (NRC,2011). Apart from a need for low levels of essential fatty acids for salm-on growth, supplied via small quantities of (the oil present in) fishmealor via fish oil, fishmeal itself is not an essential feed ingredient for aqua-culture per se, but it provides a near optimal complete feed in a conve-nient, cost-effective, and highly digestible product form (Tacon and

1 ‘High performance diets’ are defined as diets with high nutrient density, typically con-taining relatively high levels of energy and digestible protein.

Metian, 2008). However, the pattern of stagnating wild fisheries andthe fast growth of aquaculture risked over-dependence on a limitedrange of feed ingredients, especially fishmeal and fish oil, while the is-sues involved in replacement by alternative proteins and oils weresoon recognised (Sargent and Tacon, 1999; Naylor et al., 2000). For rea-sons of cost reduction and security of supply, cheaper alternative ingre-dients (e.g. soyabean meal; rapeseed oil) have therefore beenresearched and progressively substituted in commercial feed formula-tions, where technically and economically feasible, and after furtherprocessing as required (NRC, 2011). At the same time increasinglyhigher inclusion of oil has beenmade possible by extruded feed technol-ogy and driven by the protein sparing effect of oil promoting growth(NRC, 2011).

2.2. Salmon feed composition in Scotland and Norway

Information about salmon feed formulation ismore widely availablein Norway than Scotland due to reporting requirements and policy is-sues. However, until recently the same three fish feed companies dom-inated feed manufacture in both countries (i.e. BioMar, Ewos andSkretting, althoughMarineHarvest has now also entered feedmanufac-ture in Norway). Therefore, the two salmon farming industries are sim-ilar with some key differences in response to UK market focus, e.g. onomega-3. Thus feed formulation trends are broadly comparable be-tween the two countries, with Scottish salmon feeds beingmore conser-vative in terms of rate of change and more varied in reflecting bespokecustomer requirements. Fig. 1 shows the large changes in compositionof Norwegian salmon feed since 1990 when c. 90% of the diet camefrom marine ingredients compared with 29.2% in 2013 (Ytrestøyl etal., 2014, 2015).

The feed comparison between Scotland and Norway is complicatedby the proliferation of bespoke diets in Scotland, someofwhich are driv-en by external standards (e.g. Organic, Label Rouge etc.), but mostly byfarm customer requirements linked in turn to retail requirements. Al-though the consultation revealed average Scottish salmon diets for2013/14 comprised approximately 60% plant and 40% marine ingredi-ents, this obscures the large range of different Scottish formulations;for instance in 2014 c. 7000 tonnes (4%) of salmon production wasmade for export to the Label Rouge salmon specification with a marinefeed content of 51%, whereas there is also significant salmon volumebeing produced approximating to lower Norwegian diet specifications.At the same time there has been a move in Scotland during 2013/2014 towards use of high performance diets of variable specification,but generally higher use of fishmeal. The consultation also revealedthat nearly all Scottish feeds are being currently formulated to givenot b22 MJ/kg in order to achieve faster growth and improved feed

Fig. 1. Nutrient sources in Norwegian salmon farming from 1990 to 2013. Each ingredienttype is shown as its percentage of the total diet.Ytrestøyl et al. (2015).

51C.J. Shepherd et al. / Aquaculture 467 (2017) 49–62

conversion. Typically this entails diets with 37% lipid and 36% protein(with 11% carbohydrate/fibre, 9% ash and 7%moisture), with high salm-on prices allowing farmers to invest in higher specification diets athigher cost in order to yield higher marginal productivity. As a conse-quence the overall average marine protein and oil content of Scottishsalmon diets is probably around 25% and 15%, respectively, (c.f. around20% and 10%, respectively in Norway), and with higher averageeicosapentaenoic acid (EPA, 20:5n-3) and docosahexaenoic acid (DHA,22:6n-3) required for Scottish farmed salmon. In Scotland (and Nor-way) there is no use of land animal by-products, which account forover 20% dietary inclusion in Chile and Canada, or of genetically modi-fied (GM) ingredients (e.g. Chile and Canada use GM soya).

The following ingredients are commonly used in Scottish salmonfeeds: soy protein concentrate (SPC); rapeseed oil; fishmeal; wheatstarch; fish oil; wheat gluten; sunflower meal; faba beans; pea starch;maize; pea protein concentrate; and other ingredients have been usedinminor proportions, e.g. rapeseed protein concentrate, palm oil, tapio-ca starch etc., with usage based on least cost formulation in order to op-timise recipe cost. The volatility of ingredient costs is illustrated by thepattern of changing ingredient prices for five key rawmaterials in salm-on feed during the period from 2006 to 2013, with a rising trend forfishmeal, fish oil, and soya, compared with more stable prices for rape-seed oil and wheat (Fig. 2).

2.3. Effect of reduced marine ingredient inclusion on feed sustainability

The replacement of finite marine feed resources by plant-based in-gredients (and fishery by-products), is enabling salmon farming toachieve net fish protein production and possibly net marine oil produc-tion (Crampton et al., 2010; Bendiksen et al., 2011; Sanden et al., 2011).This continuing trend towards substitution of marine ingredients, to-gether with increasing evidence that reduction fisheries are not beingoverexploited, contradicts the traditional view that salmon farming re-lies on the unsustainable use of marine ingredients (Shepherd andLittle, 2014). For global ‘fed’ aquaculture, there is an overall plateau infishmeal and fish oil use despite increasing aquaculture production(Fig. 3), which has not been restricted by, at best, a static supply of ma-rine ingredients since 2009. Rising demand for fish feed ingredients hasnot therefore increased pressure on wild fish resources because of in-creased use of plant resources. At the same time there is no evidencethat plant resources are inherently more sustainable for farming salm-on, provided marine feed ingredients are sourced from sustainablymanaged stocks (or from fishery by-products); hence an increasedutilisation of plant ingredients may not be as sustainable as generally

Fig. 2. Prices of fishmeal, fish oil, rapeseed oil, soymeal and wheat delivered Europe, from2006 to 2013.Source: Marine Harvest (2014), after Chr. Holtermann.

assumed (Torrissen et al., 2011). However, all marine ingredients sup-plied to the Scottish salmon feed industry are required by the supplychains to show that they are responsibly managed and renewable (seeSection 6).

2.4. Effect of reduced marine ingredient inclusion on salmon nutrition andwelfare

If the available EPA and DHA for global salmon feed falls below 4% ofthe total dietary oil fraction (equivalent to c. 1.2% of total feed), this thenapproaches the 1% risk area for essential fatty acid deficiency in salmonand could predispose to health problems or reduced performance(Ruyter et al., 2000a, 2000b). This risk is likely to be exacerbated bythe increased use of higher performance diets and resulting highergrowth rates (NRC, 2011). The implications of falling omega-3 LC-PUFA levels for salmon consumers are considered in Section 4. In thecase of increased plant protein usage, it is well recognised that therisks relate to the need to supplement with certain amino acids (e.g. ly-sine and methionine), whereas the presence of anti-nutritional com-pounds requires prior processing of the raw materials (e.g. byconcentration or heat treatment) (Gatlin et al., 2007). As regardsmicronutrients, increased plant protein levels may lead to suboptimallevels of certain minerals (e.g. selenium, iodine) and vitamins (e.g. vita-mins A, D, and some B vitamins) (NRC, 2011).

2.5. Effect of reduced marine ingredient inclusion on contaminant levels

The Norwegian Scientific Committee for Food Safety concluded thatthe benefits of eating fish clearly outweigh the negligible risk presentedby current levels of contaminants and other known undesirable sub-stances (VKM, 2014). As regards farmed salmon, concentrations of di-oxins and dioxin-like polychlorinated biphenyls (PCBs), as well as ofmercury, have decreased by about 30% and 50%, respectively, comparedwith the corresponding levels in 2006, due to replacement of marinefeed ingredients by plant ingredients. New fish feed contaminants,such as the pesticide endosulfan, polyaromatic hydrocarbons (PAHs),and mycotoxins are unlikely to be a safety issue, since the concentra-tions are very low and not detectable with sensitive analytical methods(VKM, 2014). The health risk associated with brominated flame-retar-dants (PBDEs) is low. In contrast to their earlier conclusions, VKMstatedthat there is now no reason for specific dietary limitations on fatty fishconsumption for pregnantwomen. It is likely that the contaminants pic-ture in Scottish salmon feed closely parallels the Norwegian experience,with the main differences due to local UK sourcing of fishery by-prod-ucts, the slower rate of replacement of marine ingredients in Scottishdiets, and the greater use of bespoke salmon diets in Scotland withhigher marine content. The switch from marine to predominatelyplant-based diets will not mean a greater risk of pesticide contamina-tion of the feed, since any such contamination is unlikely to penetratethrough the hull, which is normally discarded from oil seeds duringprocessing.

3. Global supply and local demand for proteins and oils

3.1. Marine ingredients

Fishmeal and fish oil are globally traded commodities similar toother raw materials or primary agricultural products that have longbeen produced and marketed; aquaculture grew to become the biggestcustomer for these marine resources (Jackson and Shepherd, 2012).Figs. 4 and 5 detail annual world fishmeal and fish oil production, re-spectively, by producing region from 2006 to 2015, with Peru the dom-inantworld supplier followed by Chile. Theremay be a decreasing trendfor marine ingredient production (e.g. fishmeal production in 2012 and2013 was around 12% less than the 11 year mean at 4.56 and4.68 million tonnes, respectively (IFFO, 2014)). Production is subject

Fig. 3. World fishmeal and fish oil consumption by aquaculture (right axis) (—) compared with growth in ‘fed’ aquaculture (left hand axis) (___) 2000–2012 (million tonnes).Sources: IFFO data and FAO (2014).

52 C.J. Shepherd et al. / Aquaculture 467 (2017) 49–62

to environmental influences and, whereas the potential impact of cli-mate change is not well understood (Callaway et al., 2012), acute phe-nomena, such as El Niño, have well-known consequences, especiallyfor the Peruvian anchovy fishery (Shepherd and Jackson, 2013). Giventhese supply uncertainties in the dominant southern Pacific anchovyfishery, compounded by increased regulatory focus on precautionarycatch quotas, continuing growth by major users of marine ingredients(including salmon farming) has only been possible by the increasedsubstitution in feeds by other ingredients.

Based on a 2014 Scottish salmon feed volume of c. 220,000 tonnes,the corresponding estimated usage of marine ingredients at 40% offeed is 88,000 tonnes (i.e. requirements for fishmeal at 25% being55,000 tonnes and for fish oil at 15% being 33,000 tonnes). UK andIrish production of fishmeal in 2014 was c. 39,000 tonnes (T. Parker,United Fish Products, personal communication), but in practice atleast 50% of fishmeal requirements need to be imported as some ofthis domestic production is from salmon processing offal and thereforecannot be fed back to salmon. Logistics favour Scottish fishmeal importsfrom Iceland and Norway, while fish oil is also imported from Peru andChile to obtain higher levels of EPA and DHA than would be availablefrom fisheries in the North East Atlantic.

5,2865,698

5,1614,903

4,648

0

1,000

2,000

3,000

4,000

5,000

6,000

7,000

2006 2007 2008 2009 2010

Peru Chile Thailand U.S

*Includes Denmark and Norwaysource: IFFO and ISTA Mielke GmbH, OIL WORLD

Fig. 4. Annual world fishmeal production by major nat

3.2. Plant ingredients

3.2.1. Plant proteinsUS World Agricultural Supply and Demand Estimates (October

2015) have updated 2015/2016 global crop estimates, as follows:wheat - 732.8 million tonnes; corn - 988 million tonnes (based on38,898 million bushels); and soybeans - 320.5 million tonnes (USDA,2015).

The requirement for feeds with a high nutrient density ordinarily fa-vours using plant protein concentrates. These are most readily availablefrom oilseeds, including rapeseed and sunflower seed, but the mostwidely used is SPC. Although their global availability as bulk commodi-ties is not a constraint onfish feed (e.g. 2015/2016 estimated global sup-ply of oilseeds is 531 million tonnes; USDA, 2015), the availability ofprotein concentrates is less.

Norway's salmon industry requirements for non-GM soy (as SPC)were 365,000 tonnes in 2013 (Ytrestøyl et al., 2014) and current plantprotein requirements for Scotland are estimated at 77,000 tonnes,over 50,000 tonnes as SPC. Although estimated global production ofsoybeans for 2015/2106 is 320.5 million tonnes, this is mostly GMma-terial, whereas European salmon feed uses only non-GM soya products

5,854

4,743 4,9304,509

4,787

2011 2012 2013 2014 2015

.A Japan Scandinavia* Others

ional producer from 2006 to 2015 (tonnes × 1000).

9781,064 1,082 1,046

905

1,098

935 922 929846

0

200

400

600

800

1,000

1,200

2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

Peru Chile Scandinavia* U.S.A Japan Other

* Includes Denmark, Norway & Icelandsource: IFFO and ISTA Mielke GmbH, OIL WORLD

Fig. 5. Annual world fish oil production by major national producer from 2006 to 2015 (tonnes × 1000).

53C.J. Shepherd et al. / Aquaculture 467 (2017) 49–62

(see Section 6.1). Currently 5.5million tonnes of non-GM soya are beinggrown,mainly in Brazil, and supplied toworldmarkets at a price premi-um of US$80 – US$100 or c. 10% over GM soya (C. Meinich, Chr.Holtermann ANS, personal communication). Given that GM soya isused in Chilean and North American salmon feeds, requirement fornon-GM soy for Scottish salmon feed is c. 1% of global supply and totalsonly 7.5% for combined Norwegian and Scottish salmon production;however, these are underestimates as the yield of concentrates is c.61.5% of harvested soya. Despite it being a niche product and allimported mainly from Brazil, there are no indications or reasons to sus-pect that availability of non-GM soya will decline in the short to medi-um term.

3.2.2. Plant oilsPlant oils are incorporated into salmon feeds to provide energy and

blended with sufficient fish oil of the appropriate quality to give the re-quired content of EPA andDHA. The preferred ingredient is rapeseed oil,which is grown in the EU (including the UK), butmainly in North Amer-ica. Rapeseed has been cross-bred to reduce levels of erucic acid and glu-cosinolates and is safe for human consumption (e.g. Canola). Globalproduction of rapeseed oil for 2014/15 was 26.98 million metric tonsand availability of non-GM rapeseed oil is not a constraint, includingfrom UK crop material (Statista, 2015). Other plant oils, including soya

Fig. 6.World fish oil consumption (tonnes × 1000) by aquaculture, direSource: IFFO.

oil and palm oil, have been used at low levels (e.g. to help pelletintegrity).

4. Omega-3 LC-PUFA scarcity

4.1. Introduction

World usage of fish oil from 2003 to 2013 is given in Fig. 6, with in-creasing demand from direct human consumption (so-called‘nutraceuticals’) appearing to be at the expense of aquaculture. Aqua-culture use averaged around 800,000 tonnes per annum until 2012,but declined to just below700,000 tonnes per annum in 2013. Thereforearound 75% of total global supply was used in aquaculture, with 21%going for direct human consumption in 2013. Around 83% of fish oilused in aquaculture feeds in 2013 was consumed by salmonids (60%,mainly salmon) and marine fish (23%) (Fig. 7).

Although prices for fish oil and rapeseed oil were broadly similar adecade ago, demand for fish oil is now largely a derived demand foromega-3 LC-PUFA. This is shown in Fig. 2, which indicates that the fishoil price has been at an increasing premium over rapeseed oil since2009 which is directly linked to its content of EPA and DHA (Shepherdand Bachis, 2014). Thus rapeseed oil, which has no EPA and DHA, is

ct human consumption (DHC), and other uses from 2005 to 2015.

60%

4%

23%

6%0%

2%5%

Salmonids Other Marine fish Crustaceans Cyprinids Eels Tilapias

Fig. 7.Use (% of total aquaculture use) of fish oil by different categories of farmed fish andcrustaceans in 2013.Source: IFFO.

54 C.J. Shepherd et al. / Aquaculture 467 (2017) 49–62

included as a more cost-effective source of dietary oil for energy pur-poses only.

4.2. Acute shortage in 2014/2015 and likely recovery

As a result of Peru's industrial fleet being given low anchovy fishingquotas in 2014, the fleet caught only 2.25million tonnes in 2014, whichis around 1/3 of the catch in recent years (Reuters, 2015). This severelyimpacted on the global availability of EPA and DHA resulting in recordprice levels reaching US$2500–US$3000/tonne for high omega-3 Peru-vian fish oil delivered into Europe (IFFO, unpublished). Table 2a indi-cates 2014 global production was only 140,000 tonnes of combinedEPA andDHA (comparedwith a ‘normal’ year of up to c. 200,000 tonnes)giving total availability of 170,000 tonnes including year-end stockcarry-over (after Meinich, 2014). Table 2b indicates total consumptionof combined EPA andDHA for salmon feed in 2014was c. 50,000 tonnes,equating to an inclusion rate in the oil fraction of global salmon feed ofaround 6% after taking account of consumption by non-salmonid aqua-culture, direct human nutrition, and technical uses (after Meinich,2014).

Although the total Peruvian anchovy catch increased to3.69 million tonnes in 2015 (E. Bachis, IFFO, personal communication),the anchovy were leaner with a lower oil yield than 2014 due to the ElNiño effect; therefore it is considered that the current anchovy oilscarcity is continuing and global EPA and DHA supplies in 2015 orearly 2016 may be no more than in 2014 (E. Bachis, IFFO, personalcommunication).

Table 2Estimated global production (a) and consumption (b) of combined EPA andDHA (tonnes)in 2014.After Meinich (2014).

a

Estimated global production of crude fish oil 800,000Estimated global production of EPA + DHA 140,000Estimated global year-end carryover of EPA + DHA 30,000Estimated global availability of EPA + DHA 170,000

bEstimated global consumption of EPA + DHA in salmon feed 50,000Estimated global consumption of EPA + DHA by non-salmon aquaculture 50,000Estimated global consumption of EPA+DHA for direct human consumption 61,000Estimated technical usage of EPA + DHA (e.g. hardening, tanning, energy) 9000

Due to high fish oil prices, non-traditional sources are nowappearing on world markets, e.g. Sardinella spp. from north-west Africa(C. Meinich, Chr. Holtermann ANS, personal communication), althoughsuch sources are unlikely to be satisfactory in regard to certification re-quirements for UK salmon feed use (see Section 5.1). High prices willtend to dampen demand from price sensitive segments. Demandgrowth for human nutrition and non-salmon aquaculture would be ex-pected to be around 2.5% - 5% per annum; however, human nutritiongrowth has stalled since 2013 (A. Ismail, GOED, personal communica-tion) and fish oil inclusion in non-salmonid feeds can be expected tofall at current prices.

Historical precedent provides confidence of a likely strong reboundin Peruvian anchovy stocks following the current oceanic conditions ofEl Niño, with a recovery now likely during 2016, in which case annualglobal fish oil supply will then increase towards more ‘normal’ annualproduction levels approaching c. 1 million tonnes of crude fish oil (cor-responding to up to amaximumof 200,000 tonnes of combined EPA andDHA). History also suggests that such ‘normal’ levels will be sustaineduntil the next El Niño event, which is unlikely to occur before 2020. As-suming a 2.5% compound annual growth rate (CAGR) for both directhuman nutrition and salmon farming and 5% CAGR for non-salmonaquaculture from 2016, this will permit pro-rata growth in EPA andDHA demand, with total consumption of 170,000 tonnes in 2016 in-creasing to 192,000 tonnes in 2020 (including 55,000 tonnes for salmonbut with spare capacity to increase to 63,000 tonnes). If another El Niñooccurs soon after 2020, by then the resulting scarcity could possibly beoffset by an increased supply of EPA and DHA from novel sources (seeSection 4.6).

4.3. Market significance of changing EPA and DHA content of salmon

Onozaka et al. (2012) surveyed the position of farmed salmon in fiveEuropean countries (including the UK) and found that consumers re-gard salmon as superior in healthiness compared to other meats. Cer-tain UK retailers reported that their customers take health benefits ofsalmon into account in their purchasing decisions, hence negative pub-licity on the health benefits of farmed salmon might reduce demand.However, the health benefits of farmed salmon consumption are not re-stricted to its content of omega-3 LC-PUFA (Galli and Risé, 2009; Tocher,2009). Salmon comprises highly digestible protein and essential aminoacids and marine lipids, as well as vitamins and minerals, so consump-tion of salmon is more nutritious compared with consuming omega-3supplements alone.

The Global Organisation for EPA and DHA (GOED) summarized theinternational recommendations for EPA and DHA intake (GOED,2014), while Shepherd and Bachis (2014) showed that, as a result ofthe combined EPA and DHA content of added oil in Norwegian feedsfalling from 20% in 2002 to 7.2% in 2012, the number of days recom-mended requirementmet by one portion of Norwegian salmonhad fall-en proportionately to approximately 6 days (EFSA at 250 mg/day),3.4 days (SACN at 450 mg/day), and 3 days (ISSFAL at 500 mg/day).The average UK consumer eats less fish than Norwegians and a UK sur-vey showed that, for oily fish, average consumption waswell below therecommended one (140 g) portion perweek in all age groups,withmenandwomen aged 19 to 64 eating just 52 g and 54 g/week of oily fish, re-spectively (HM Government, 2014).

In a recent study of different farmed andwild salmon products avail-able during 2013 in retail outlets in Scotland, Henriques et al. (2014)showed that, although wild salmon products had higher relative valuesof EPA and DHA, farmed salmon products generally delivered a higherdose of EPA and DHA compared to the wild salmon products, due totheir higher lipid content, and were therefore better able to deliver rec-ommended dietary intake levels than the wild salmon products; thestudy also confirmed the elevated levels of omega-6 PUFA, specificallylinoleic acid (18:2n-6), but concluded that 18:2n-6 does not have a

55C.J. Shepherd et al. / Aquaculture 467 (2017) 49–62

major impact on the nutritional quality of farmed salmon and does notoutweigh the benefits of the high omega-3 LC-PUFA levels recorded.

4.4. Omega-3 LC-PUFA differentiation by UK retailers

The consultation revealed there is clear segmentation in UK farmedsalmon supply between those retailers willing to claim that one portionprovides the weekly required intake (‘formulated to deliver high levelsof healthy omega-3 long-chain fatty acids’ or similar wording) andthose simply stating ‘variable omega-3 content’ or similar wording. Cur-rently four (out of nine) UK retailers (so-called ‘premium suppliers’)claim on the pack of some or all of their products that consuming thestated portions will deliver the internationally recommended intake ofEPA andDHA, forwhich EFSAwould appear to be the (minimum) targetreference of choice. For these premium suppliers/products, there is aclear link being made between the content of omega-3 LC-PUFA onoffer and the health of the consumer, with the premium suppliers spec-ifying Scottish farmed provenance, i.e. omega-3 has become somethingof a Unique Selling Point (USP) linked to Scottish production. This differ-entiation is used as a competitive tool by retailers, but raises relatedquestions, such as:

• The term ‘omega-3 fatty acids’may justifiably include more than EPAand DHA, e.g. docosapentaenoic acid (DPA). In evaluating competingproduct claims, care must be taken to ensure like-for-like compari-sons, since using total omega-3 LC-PUFA content gives higher fleshvalues than only EPA and DHA.

• There is no indication whether premium salmon is formulated withan overage of omega-3 LC-PUFA to take account of the reduction inlevels due to cooking the product (which will vary with cookingmethod and fat loss, etc.), while intake recommendation levels arebased on an assumption of what is actually consumed.

• There is no reference to omega-6 fatty acid levels that have increasedin farmed salmon due to vegetable oil inclusion levels.

4.5. The challenge to feed formulation

The existence of a range of bespoke salmon products in response toretail market differentiation represents a complex challenge for farmsand feed suppliers. From the consultation it appears that UK feed millsadd between 6.5% and 8.5% of combined EPA and DHA in the oil fractionof the feed to achieve 1.75 g in a 130 g fillet portion of salmon (EFSA tar-get), depending on salmon bodyweight and fat level (or proportionatelymore assuming a 100 g portion). The lower end of the Scottish farmedrange is in the normal range of standard Norwegian practice during2014 and is claimed to be typical of current UK imports of Norwegiansalmon (supplying an estimated 60% of the UK market). At theselower levels there is no certainty that the EFSA claimwill bemet consis-tently (unless consumers eat more salmon each week) and this lowerrange corresponds to the majority of UK (non-premium) salmon vol-ume labelled on the basis of ‘variable omega-3 levels’.

4.6. Future availability of LC-PUFA sources

GM sources of LC-PUFA (EPA and DHA), such as EPA-containing GMyeast, are being used in Chilean salmon feeds; ‘Verlasso salmon’ (http://www.verlasso.com/). In addition, commercial production of EPA andDHA from GM oilseeds could be a reality by around 2020 (Section8.2.2) (Tocher, 2015). Other potential sources of LC-PUFA are thefollowing.

4.6.1. Fishery and aquaculture by-productsIncreasing use of fishing industry by-products will continue to sup-

ply marine ingredients, including fish oil. The recycling of salmon oilfor salmon production, which is prohibited by some codes of practice,

is being used in aquaculture feeds (e.g. for gilthead bream, Sparusaurata) and could potentially supply c. 150,000 t of global salmon oil.With 70% vegetable origin it would contain a maximum 10,000 t ofEPA and DHA, but its use would reduce the total demand from non-sal-monid aquaculture, making more EPA and DHA available for salmon. Inconclusion, increased EPA and DHA from fishery and aquaculturesources is unlikely to havemajor volume significance in the short tome-dium term (Tocher, 2015).

4.6.2. Whole cell DHA-rich algal biomass‘DHA Gold’ (http://www.dhagold.com/) is a dry biomass sold by

DSM as a non-GM feed product approved for sale in the EU.

4.6.3. Microalgal DHAFermentation of heterotrophic organisms (e.g. Schizochytrium)

(Tocher, 2015) requires fermentation use of energy-dependent organicsubstrates (e.g. sugar), with current commercial developments by DSM,ADM, Alltech, etc. The potential applicability of DHA to salmon feed is byblending fish oil to achieve a similar 1:1 ratio of EPA to DHA as seen inwild Atlantic salmon. As Peruvian anchovy has an 18:12 ratio of EPAto DHA, DHA can be added to anchovy oil to achieve a 1:1 EPA to DHAratio, before balancing the blend with rapeseed oil; hence 25–30% ofthe resulting ‘fish oil’ can come from microalgal DHA but still remainin the natural range of fish oil composition of wild Atlantic salmon.

4.6.4. Longer term optionsIn the long term it is recognised that cultivation of autotrophic/

phototrophic algae using photosynthesis is the most efficient solution,but it is proving highly complex and difficult to scale up (Tocher, 2015).

4.7. Options and implications for the Scottish salmon industry

The current shortage of marine sources of LC-PUFA (EPA and DHA)(Section 4.2) should start to reverse soon and supplies are likely to be-come normal during 2016 and thereafter until the next El Niño occursaround 2020, by when alternative novel sources of EPA and DHA maybe available from GM and non-GM sources; otherwise the currentacute shortage becomes a chronic shortage. Total 2015 supply of fishoil (including farmed salmon oil) is estimated at 834,800 tonnes (E.Bachis, IFFO, personal communication) and the expectation is that aver-age salmon feed inclusionwill have remained close to 6% combined EPAand DHA level in feed oil, corresponding to the bare minimum neededto meet EFSA weekly intake recommendation. The main difficulty forthe Scottish industry from the current shortage is if the premium seg-ment needs to reassess its claim about one portion of fish meetingweekly recommended requirements for omega-3 LC-PUFA; if productrelabelling becomes necessary, this could negatively impact salmon'simage as a healthy product.

It is relevant for the industry to consider the marginal cost of addi-tional EPA and DHA in Scottish feed. For example, assuming the replace-ment cost of anchovy oil (c. 26% combined EPA and DHA) is US$3000/tonne and it is necessary to add 7% of the oil fraction of the feed asEPA and DHA to reliably meet the EFSA claim, this is equivalent at 30%oil inclusion to 2.1% of the total feed, hence costing US$240/tonne. Onthis basis should the market be supplying only 5% of the oil fraction ofthe feed to conserve material, the additional 2% to regain scope for theEFSA claimwould cost c. US$69/tonne (or US$34.50 for each percent in-clusion of combined EPA and DHA), which may be viewed as a minorcost premium set against the reputational cost and risks of doing other-wise. On the samebasis, if current salmon feed cost/tonne is c. US$1250/tonne, an additional 2% of combined EPA and DHA at US$69/tonnewould increase feed cost by 5.5%. Note (i) calculations should takeinto account the inclusion of fishmeal, which contains 9% oil with highlevels of EPA and DHA; (ii) if some producers are using more EPA andDHA, others will have to use less due to scarcity limits.

56 C.J. Shepherd et al. / Aquaculture 467 (2017) 49–62

5. Role of certification and standards in responsible sourcing

5.1. Demonstrating responsible sourcing of marine feed ingredients

Standards and certification schemes have become important tools tomanage concern issues, such as economic, social and environmentalsustainability (e.g. Gail Smith, 2008). Themain focus for Scottish salmonis long-term sustainability from an environmental standpoint, especial-ly use ofmarine feed ingredients. An overviewof the sustainability of re-duction fisheries is published annually (Sustainable FisheriesPartnership, 2015) and there are well-established food chain standardscontrolling process quality, food safety, and a particular focus on fishwelfare in Scotland (Munro and Wallace, 2015).

In developing a consensus approach to responsible practice, theCode of Good Practice for Scottish finfish aquaculture (CoGP) seeks ‘toenhance the industry's reputation for respecting the environmentthrough adoption of best practice and greener technologies and reduc-ing the impact on wild fisheries by increasing use of alternative feedsources’ (CoGP, 2015). In addition to recommendations on feed formu-lation and use, the CoGP requires that fish-catch supplies used infishmeal manufacture are from fisheries which are responsibly man-aged, either by reference to the FAO (Food and Agriculture Organisationof the United Nations) Code of Conduct for Responsible Fisheries(FAOCRF; FAO, 2015) or the Marine Ingredients Organisation (IFFO)RS scheme (IFFO, 2015), or by another globally recognised standard.The CoGP also makes fish welfare provisions, which are influenced byhusbandry conditions, including use of suitable feeds and feedingmethods.

To demonstrate responsible sourcing, the use of certified ingredientshas been adopted by Scottish feed suppliers and farmers (SSPO, 2013).Formarine feed ingredients used in aquaculture there are six commonlyused standards; GlobalGap is more focused on food safety, while theother five claim to be based on the key principles underlying theFAOCRF (FAO, 2015). They are Aquaculture Stewardship Council (ASC)(http://www.asc-aqua.org), Best Aquaculture Practice (GAA BAP)(http://www.gaalliance.org/bap/standards.php), Friend of the Sea(http://www.friendofthesea.org/), IFFO RS, and Marine StewardshipCouncil (MSC) (http://www.msc.org). For marine ingredients, IFFO RScertification is probably universal in Scotland providing evidence oftraceability back to responsibly managed fish stocks, avoidance of ille-gal, unreported, and unregulated fish (IUU), and control of by-productraw material, with an associated chain of custody. For sourcing offishmeal and fish oil, the following were required for the processorsand retailers interviewed during this study:

• Traceability to species and country of origin• No endangered species used, as defined by the International Union forConservation of Nature (IUCN) ‘Red List’ (IUCN, 2015)

• Preference for feed manufacturers to provide evidence of responsiblesourcing

• Avoidance of IUU fish

5.2. Demonstrating responsible sourcing of plant feed ingredients

Three standards are currently used for plant ingredients for Scottishsalmon feeds:

• The Roundtable on Responsible Soy (RTRS) Association approved itsStandard for Responsible Soy Production versions in 2010 (RTRS,2011)

• The new ProTerra Foundation Standard Version became effective inJanuary 2015 and includes soya (ProTerra, 2015).

• Cert ID is focused on Non-GMO certification and has become thebenchmark for Non-GMO identity preservation (Cert ID Europe Ltd.,2015).

The use of these three schemes for fish feed, primarily for soya, isseen aswork in progress. Given the growing recognition of potential en-vironmental problems (e.g. rain forest degradation associated with un-controlled soya production, especially in South America), it is perhapssurprising that environmentalist pressure has focused on use of marineingredients in aquaculture and largely ignored plant ingredients untilrecently.

5.3. UK supply chain perspective - salmon farmers and feed manufacturers

Itwas clear from the consultation that sustainability issues are a stra-tegic concern for the owners of the three principal UK salmon feed com-panies and lie at the heart of their day-to-day operations, including rawmaterials purchase; also that salmon purchasing firms increasinglyoblige their suppliers to provide evidence demonstrating responsiblemanagement and use of renewable resources. The salmon farm andfeed-related certifications used currently by Scottish salmon farms andtheir associated supply chains are given in Supplementary File 1. Keyfindings include the following.

(i) Over 90% of Scottish salmon farms are members of the ScottishSalmon Producers Organisation (SSPO) and subscribe to theCoGP.

(ii) Widespread adoption of ‘Freedom Food’/‘RSPCA Assured’ certifi-cation (http://www.freedomfood.co.uk/) is an internationaldifferentiator specifying ‘sustainable feed’ and IFFO RS certifica-tion of marine ingredients

(iii) The most widely (but not exclusively) used Scottish salmonfarming certification is GlobalGap (http://www.globalgap.org/uk_en/), which is focused on food safety and feed assurance.

(iv) The ASC salmon standard (http://www.asc-aqua.org/index.cfm?lng=1) has received global endorsement byMarineHarvestwithtwo Scottish salmon farms now certified. ASC is likely to beadopted by other Scottish producers

(v) IFFO RS certification is currently only accepted in a chain of cus-tody role by the ASC salmon standard.

(vi) The GAA BAP certification is not currently adopted by any Scot-tish producers despite its leading position in North and SouthAmerican aquaculture (http://gaalliance.org/what-we-do/bap-certification/).

(vii) Greater harmonisation between different standards would bebeneficial; the Global Sustainable Seafood Initiative plans tobenchmark them to facilitate sourcing decisions (http://sustainableseafoodcoalition.org/news/new-gssi-website-launches).

5.4. UK supply chain perspective - salmon processors and retailers

To discover salmon sourcing policies of leading multiple retailersand processors, eight of the nine major UK retailers and two leadingUK seafood processors were consulted. This confirmed that UK retailsalmon standards are a complex set of competing codes aimed at in-creasing retailers' control over the supply chain. Aquaculture focus onsustainable feed issues is partly driven by competing UK retailers andis influenced by pressure from environmental NGOs, which often com-pete for support from consumers and supply chains. The existence ofdifferent private standards can cause confusion (e.g. to retail fishbuyers) and is poorly understood by consumers. It appears sustainabil-ity issues are being used as a competitive tool by firms highlighting re-sponsible sourcing credentials, so salmon farming and feed standardscontinue to be driven up; however, this has implications in restrictingchoice among different raw material sources.

57C.J. Shepherd et al. / Aquaculture 467 (2017) 49–62

5.5. Relevance of input/output indices for marine resource use

Although the main driver for replacement of marine ingredients byplant raw materials has been the combination of nutritional innovationand market forces in the face of fluctuating supply and highly volatileprice, replacement has impacted on the different indicators used to relatemarine ingredient input to farmed salmon output. These ‘marine sustain-ability indicators’ include fish-in fish-out ratio (FIFO), marine protein de-pendency ratio (MPDR), forage fish dependency ratio (FFDR), etc., andtheir use has been reviewed by various authors (e.g. Ytrestøyl et al.,2015). Although FIFO and FFDR have been used as a proxy for sustainabil-ity, it is not clear why this should be since sustainabilitymust be based onresponsible harvesting of fish that are used for fishmeal and fish oil ac-cording to international regulations (backed up by private standards ifnecessary). Nor has the FIFO ratio any obvious nutritional basis and it isnot therefore a measure of production efficiency (Tocher, 2015).

As one of its indicators ASC has included FFDR, calculated as thequantity of forage fish required to produce the amount of fishmealand fish oil to produce a unit of farmed fish. For 2013 the correspondingFFDRs for Norwegian salmon were 1.54 and 0.69 for fish oil and meal,respectively, well within the ASC standards of b2.95 for fish oil andb1.35 for fishmeal (Ytrestøyl et al., 2014); by comparison the equivalent(unpublished) values for one Scottish producer in 2013 of 2.09 for fishoil and 0.76 for fishmeal were higher than Norway, but well withinthe ASC permitted range.

5.6. Effect of changing raw materials on salmon farming sustainability

As regards protein retention, Torstensen et al. (2008) and Bendiksenet al. (2011) have found similar levels inAtlantic salmon fed eithermainlymarine diets or mainly plant-based diets, explaining why salmon can beproduced with feeds containing high inclusions of plant ingredients andonly low inclusions ofmarine ingredients. Changing the salmondiet com-position from88%marine ingredients to 85% plant ingredients resulted inalmost the same carbon footprint (Ytrestøyl et al., 2011), which supportsthe viewof Torrissen et al. (2011) that increased utilisation of plant ingre-dients may not provide an increase in sustainability as often claimed. Theincreased use of fishery by-products influences sustainability; althoughthey have little, if any, alternative use, the effect on dietary carbon foot-print will depend on whether the life cycle analysis (LCA) calculationadopts economic or mass allocation. As stated by Ytrestøyl et al. (2015),several methods must be used to evaluate eco-efficiency and sustainabil-ity of salmon production as no single method is sufficiently robust.

6. Scope for alternative feed ingredients, including land animalproducts, genetically modified, and fermentation products

Considerable research has been performed to assess a large range ofalternative protein and oil/fat sources as ingredients for aquaculturefeeds, including salmon (Gatlin et al., 2007; Turchini et al., 2011).Focus here will be on alternative current and emerging protein and oilingredients that have large-scale potential in Scottish salmon farming,but where there are supply chain concerns about their use or the tech-nical basis is not sufficiently established. For alternative protein feed in-gredients found to be suitable and cost-effective, it is important to beable to concentrate the protein in order for it to replace SPC in practicalfeed manufacture.

6.1. Land animal by-products (LAPs)

6.1.1. Current status and availability of LAPs in EuropeBefore year 2000 animal proteinswere widely used in fish feeds, but

due to Bovine Spongiform Encephalopathy (BSE), most animal proteinswere banned from terrestrial and aquatic animal feeds. From June 2013,due to improved testing methods, use of non-ruminant Processed Ani-mal Protein (PAP) has been approved for use in aquaculture in the EU

(Regulation 56/2013). The following are the main non-ruminant PAPproducts in the UK and EU now legally available for inclusion in aqua-culture feeds: Poultry PAP; Feather meal PAP; Porcine PAP; Porcineblood meal PAP; and Porcine blood products. EU production of poul-try-derived PAP and of porcine PAP and blood products in 2013 was365,000 tonnes and 275,000 tonnes, respectively (Woodgate, 2014).To ensure lack of ruminant protein contamination, segregation, security,and traceability of ‘Category 3’ animal by-products is the subject of de-tailed controls at the slaughterhouse and downstream.

6.1.2. Current use of LAPs in salmon feedDespite the legal relaxation, since 2001 no LAPs have been used in

salmon feeds in Europe, with a marked reluctance by supply chains toincorporate animal by-products in feed for farmed salmon, despite tech-nical benefits (e.g. blood meal's histidine content preventing cataracts).The situation outside Europe with fewer BSE-related problems is verydifferent (e.g. current Chilean salmon diet formulations with 3% poultryoil and 19% LAPs).

A voluntary ban on use of LAPs in salmon feed in UK (andNorway) isreinforced by the CoGP prohibiting use of LAPs in Scottish salmon. Thisstrict policy appears contradictory in view of retail supply of importedfarmed warm water prawns (Penaeus spp.), ‘River Cobbler’ (Pangasiusspp.), and Pacific salmon (Oncorhynchus spp.) from enhancementprogrammes, which may all have been fed LAPs. The rationale of theban is fear of negative consumer reaction, variously linked to: BSE con-cerns, the horsemeat contamination issue, the supposed ‘unnaturalness’of such ingredients, and associations of animal by-products with use ofwaste material otherwise destined for landfill disposal. The retailersconsulted pointed out this is not a food safety issue and would give ac-cess to global salmon production, but accept their customers do notwish the policy to change. All respondents referred to the constraintsof halal and kosher requirements, which prohibit use of porcinemateri-al (e.g. pig blood), hence further supply chain segregation. In practicethis means that use could only be made of poultry products, such aspoultry offal meal, feather meal and poultry oil. The non-governmentalorganisations (NGOs) consulted supported the use of LAPs in salmonfeed as representing a sustainable solution; also theASC salmon farmingstandard supports the use of LAPs.

6.1.3. Conclusions and implications for LAPs/PAPConsultation, especially with retailers, showed strong resistance to

using LAPs in salmon feed by UK and continental European consumers.It is therefore not under consideration in Scotland and this is unlikely tochange in the foreseeable future, despite the EU having removed legalobstacles for PAP use. If this is reconsidered in the future, robust sourceassurance guarantees would be required and porcine material wouldneed to be excluded. As regards cost savings from an open salmonfeed formulationwith poultry by-products, estimates from feed compa-nies consulted were in the range of £60–£80/tonne (i.e. up to 10% offeed cost), depending on feed size and specification.

6.2. Use of genetically modified feed ingredients

6.2.1. Scope for GM protein ingredients

6.2.1.1. UK use in land animal feeds versus salmon feeds. Following supplyconcerns about non-GMsoybeans in 2013, all but one of the eight UK re-tailers consulted changed their sourcing policies to permit the inclusionof GM plant ingredients in terrestrial livestock feeds, especially forchicken. Although inclusion of approved GM feed ingredients is legallypermissible in the UK, currently salmon producers and feed companieshave a strict non-GM policy. This appears to reflect the following:

(i) Negative consumer attitudes to GM in Europe, especially inFrance, Germany, Austria and Italy, with potentially severe ex-port implications for Scottish salmon.

58 C.J. Shepherd et al. / Aquaculture 467 (2017) 49–62

(ii) The current lack of a clear commercial disadvantage to sourcingnon-GM ingredients, such as reduced availability and substan-tially increased cost for non-GM SPC.

(iii) The current consensus to maintaining non-GM fed salmon statusin Scotland being shared by the SSPO, some NGOs, and the Scot-tish Government, as part of a ‘green and clean’ stance (while En-glish and Welsh Government policies on GM are more flexible).

(iv) The Norwegian salmon farming industry has a non-GM policy.

However, seven of the eight UK retail respondents consulted werewilling to consider adopting a more flexible GM feed policy if marketcircumstances changed. They accepted that UK consumer attitudeswere becoming less hostile to GM, but the largest supply chain concernrelates to the risk of losing key export markets, especially France.

6.2.1.2. Availability of non-GM proteins. As shown in Section 3.2.1, al-though the 5.5 million tonnes of non-GM soy protein is b3% of globalsoya supplies, dominated by GMmaterial, there is no current threat toavailability, or sufficient of a price driver for switching to GM SPC insalmon feeds. Discussions inNorway aboutwhether it is logistically pos-sible to supply GMand non-GM feed at the same time by using differentmills concluded this is more a matter of policy and additional cost thanlogistics, but it would be more difficult (but possible) to segregate theresulting fish streams post-harvest. This is not seen as being feasible inScotland and any move to abandon the non-GM policy would need tobe agreed jointly by all three UK feed producers.

6.2.2. Scope for GM oil containing EPA and DHACommercial production of so-called ‘Verlasso’ salmon in Chile in-

volves the use of yeast, which has been genetically modified to produceEPA (Xue et al., 2013). The transgenic yeast cells are produced by fer-mentation using glucose and killed before the whole dead cells areadded to salmon feed as a partial replacement for fish oil (Hatlen etal., 2012; Berge et al., 2013).

Camelina oil (from the oilseed crop False Flax, Camelina sativa) hasbeen suggested as commercially suitable for salmon feeds since it con-tains about 30%α-linolenic acid (a precursor of EPA and DHA), with rel-atively low levels of omega-6 PUFA, and has been used experimentallyto replace fish oil in salmon diets (Hixson et al., 2014). However, it isthe recent success in producing EPA and DHA at levels equivalent tothose in fish oil by inserting algal DNA into Camelina that has arousedmost interest (Ruiz-Lopez et al., 2014) (http://www.rothamsted.ac.uk/camelina). It has been estimated that 1 ha of the new transgenic cropwould produce about 750 kg of oil containing around 12% EPA and 8%DHA and it should be highly scalable, hence 1 million hectares couldproduce 750,000 tonnes of oil or 150,000 tonnes of EPA and DHA(Ruiz-Lopez et al., 2014). Oil from transgenic Camelina has been dem-onstrated to successfully replace fish oil in feeds for Atlantic salmon(Betancor et al., 2015a, 2015b). Although the European Parliamentagreed in January 2015 to allow individual member states to determinetheir GM policy, regulatory approval for growing the transgenic cropwould be simpler outside the EU (e.g. in Canada or USA). Transgenicoil might become available in the UK (and Norway via its EU agree-ments) by around 2020, subject to commercial agreements and to reg-ulatory approval in the country where the crop is grown and in the EUas a GM feed additive. A similar timescale seems likely for the produc-tion of EPA and DHA from transgenic canola (rapeseed) by a joint ven-ture between BASF and Cargill. In Australia the focus on transgeniccanola is mainly on DHA (Kitessa et al., 2014) with claims that commer-cial supply could be available by 2017 (CSIRO, 2013). The existence of anentire logistical system for handling oilseed products means that thesupply chain is already present and GM oil's production cost need beno higher than conventional oilseeds. The continuing decline in EPAand DHA levels is a potential driver for change if GM oil becomes

commercially available and the lack of any cellular material, protein,or nucleic acid in oil may help market acceptance.

6.2.3. Conclusions and implications for GM feed ingredientsUnless non-GMprotein becomes less competitive in terms of supply

and price, given the current premium for non-GMSPC (used at 20%–30%of the diet) is under 10%, there is little commercial incentive for theScottish industry to change policy today. The position with omega-3LC-PUFA is different as the potential commercial availability of GM oil-seed crop sources by around 2020 offers a potential technical solutionto a chronic shortage, which might otherwise undermine the healthyprofile of Scottish farmed salmon. A switch to using GM oil might beseen as worthwhile to avoid the risk of fish oil supply interruptions, ifand when transgenic oils become commercially available. If the nextEl Niño causes a repeat of the current acute scarcity, and consumer atti-tudes continue to soften, itmight become appropriate for policy-makersand the industry in Scotland to review current policy, but the potentialrisk to salmon exportmarkets, especially France, fromusingGM feed in-gredients is likely to be a key commercial consideration.

6.3. Use of novel non-GM feed ingredients

6.3.1. Insect-based feed ingredientsSince wild salmon eat insects during the freshwater stage, there is

current R&D investment on the potential of using insects as safe andhealthy ingredients for salmon feed. Makkar et al. (2014) reviewedexisting research on five major insect species that are claimed to havepotential for animal feed concluding that black soldier fly larvae havemost promise for replacing soybean meal in pig and poultry diets. Therecent ‘Aquafly’ project in Norway is exploring the potential to tailor in-sect nutrient profile to meet salmon nutritional requirements (http://nifes.no/en/counting-insects-future-fish-feeds/). It is claimed thatmany insect species are highly nutritious and their production has lessenvironmental impact comparedwith traditional sources of animal pro-tein. At an EU level, EFSA has been commissioned to review availablesafety evidence around insect protein, while the Commission is fundingthe PROteINSECT project to investigate quality, safety, processes, andhuman acceptance around the use of insects in animal feed (http://www.proteinsect.eu). In addition to salmon nutritional studies, thereis a need for cost-effectivemass insect-rearing facilities and a regulatoryframework and sanitary procedures for the safe use of bio-wastes (in-cluding managing the risks of diseases, heavy metals and pesticides,etc.). The scope for insect use as a potential protein replacement insalmon feeds is clearly at an early stage of evaluation.

6.3.2. Fermentation productsAs an alternative to DHA-rich algal biomass produced by fermenta-

tion of non-GMheterotrophic organisms, the current low cost of energyhas renewed interest in natural proprietary fermentation as a means ofproducing microbial protein based on methane as the main source ofcarbon and energy. In particular Methylococcus capsulatus, a naturallyoccurring single cell organism, is now being used to produce a stablepowder or dry pelleted product with 71% protein, 10% fat, and b1%fibre, and a shelf life of over a year. The resulting product (‘FeedKind’™)is already approved for use in the EU and it is claimed that it will be re-leased commercially in 2018 and prove a superior alternative tofishmeal and soy in aquaculture diets (http://calystanutrition.com/feedkind-protein/). Assuming the product proves cost-effective in use,this has some clear advantages to conventional protein sources in salm-on feeds. It remains to be seen if theproductwill remain feasible in prac-tice if and when energy costs escalate once more, although themanufacturers claim they could switch to alternative feedstocks, suchas biogas.

59C.J. Shepherd et al. / Aquaculture 467 (2017) 49–62

7. Projected forward requirements for salmon feed ingredients

7.1. Future farming and feed requirements

The Scottish Government's farmed salmon target of 210,000 tonnesby2020, (see Section1.2) represents a realistic CAGRof c. 2.5%, providednew production sites are available, etc. This is also in line with globalsalmon projections where annual growth rate has recently diminished,resulting in a projected 3% CAGR from 2013 to 2020 (Kontali, 2013). Ifthe overall FCR in 2013/2014 was c. 1.3 (see Section 1.2), and annualsalmon harvest volume is expected to increase from c. 165,000 tonnesto 210,000 tonnes in 2020, it seems not unreasonable to assume an im-proved FCR within the range of 1.1–1.2, corresponding to a 2020 Scot-tish salmon feed requirement in the range of 231,000–252,000 tonnes.

7.2. Forward supply of marine ingredients

OECD-FAO (2013) constructed an annual global forecast from 2013to 2022 of total fishmeal and fish oil production, production fromwhole fish, consumption, variation in stocks, and price. Using theOECD-FAO fish model for 2022, FAO (2014) projected total fishery pro-duction, aquaculture, fishmeal and fish oil production assuming thatabout 16% of capture fishery production will be reduced to fishmealand fish oil (down 7% on the 2010–2012 average, the base period),but total ‘baseline’ 2022 production will be 7.02 million tonnes and1.08million tonnes, respectively (i.e. up 15% and10%on thebase period,with almost 95% of the additional gain for fishmeal coming from im-proved use of fish waste, cuttings and trimmings) (Table 3). OECD-FAO estimate that fishmeal from by-products should represent 49% oftotal fishmeal production in 2022 and that, with global demand stron-ger than supply, prices of fishmeal and fish oil will increase by 6% and23%, respectively, in nominal terms by 2022 (OECD-FAO, 2013). Al-though the OECD-FAO model is the best available, this is no guaranteeof its predictive ability, especially if there is a possibility that the baseperiod may be somewhat overstated. For instance IFFO data givelower estimates for the 2010–2012 base period than FAO with4.92 million tonnes of fishmeal as the 2010–2012 year average versus6.10million tonnes from FAO (duemainly to lower assumptions on Chi-nese production by IFFO), while IFFO data for fish oil is close at0.96million tonnes versus 0.98million tonnes for the same 3-year aver-age (IFFO, 2014). Also IFFO data are suggestive of a recent downwardtrend in fishmeal production most likely linked to more precautionaryfishing (Fig. 4). Although this is less evident in the data on fish oil pro-duction, the trendwill be for less fish oil being available for aquaculturedue to competition from the human nutrition sector (Figs. 5 and 6).

Supply and demand forecasts for fishmeal and fish oil need to takeinto account inter alia demand for pelagic fish for direct human con-sumption, the effect on reduction fisheries of increased regulatorycurbs, climatic events (e.g. El Niño), increased exploitation of fish pro-cessing wastes for reduction purposes, etc. At the same time demandfor fishmeal and fish oil is a function of price and availability of alterna-tive proteins and alternative oils, particularly novel sources of omega-3

Table 3FAO fish model: overall trends to 2022 for world (million tonnes in live weight equivalent).FAO (2014).

Base period 202

2010–2012 Bas

Total fishery production (world) 153.940 181Aquaculture 62.924 85.1Capture 91.016 95.9Fishmeal production (product weight) 6.103 7.02Fish oil production (product weight) 0.980 1.07Fish trade for human consumption 36.994 45.0Fish supply for human consumption 131.741 160Per capita apparent fish consumption (kg) 18.9 20.7

LC-PUFA, as well as the rate of growth, not only of fed aquaculture spe-cies, but also of young pigs and day-old chicks (Shepherd and Jackson,2013). This is in turn influenced by innovation of feed formulators,plant protein processors, and geneticists, as they seek to reducefishmeal inclusion rates to save on scarce, fluctuating, and costly ingre-dients, and also by the efforts of civic society (e.g. NGOs) to discourageuse of marine ingredients on alleged sustainability grounds. Since fishoil demand has become a derived demand for EPA and DHA, salmonfeed buyers compete strongly with the human nutrition market, whilesupply is linked to that of its co-product fishmeal, with both productssubject to the volatility of the dominant Peruvian catch due to El Niño.However, Olsen and Hasan (2012) concluded that the limited supplyof pelagic fishmeal will not be a major obstacle for a continued moder-ate growth in global aquaculture production; it is reasonable to makethe same conclusion in regard to global salmon production.

On the somewhat conservative assumption of a 30% marine/70%plant content of feed ingredients (lower in marine content than today'sScottish average, but the same as current Norwegian feeds), 231,000–252,000 tonnes of salmon feed for Scotland in 2020 would require an-nually 69,000–76,000 tonnes of marine ingredients and 162,000–176,000 tonnes of plant ingredients. On this basis the reduced fishmealinclusion rate projected for 2020 will more than offset the increasedfeed production. If the UK annual fishmeal requirement in 2020 is45,000–50,000 tonnes (compared with c. 55,000 tonnes demand and39,000 tonnes supply in 2014, − see Section 3.1) and domestic supplyis likely to rise over time, most, if not all, fishmeal requirements couldbe sourced locally if desired. However, salmon by-products could notbe recycled (3.1) and sourcing criteria for farm certification (e.g. byASC), may become more exacting over time (5.2), restricting rawmaterial.

7.3. Forward plant protein and oil supply

OECD and FAO secretariats analysed the price ratios of aquaculturespecies relative to oilseed products and concluded that tight suppliesof fishmeal and fish oil are likely to contribute to an increased priceratio between fish and oilseed products over the medium term due tocontinuing demand from early rearing of pigs and salmon farming andfrom continuing omega-3 demand (OECD-FAO, 2013). They recognisedthe price ratioswill be exacerbated in El Niño years, further constrainingsupply and supporting higher prices. Also they project a 26% increase inworld production of oilseeds and a switch in distribution of land fromcoarse grains to oilseeds. Global protein meal output is projected to in-crease by 25% or 67million tonnes by 2022; this envisages consumptiongrowth of protein meal slowing somewhat due to slower absolutegrowth in global livestock production and slower growth in the relativeuse of protein meal in feed rations. The overall availability and price ofterrestrial ingredients will also depend on factors such as freshwateravailability.

In the short-term, demand from Scottish and global salmon produc-tion for the principal plant feed ingredient, non-GM soy (as SPC) is onlyc. 1% or 7.5%, respectively, of global supply, suggesting security of supply

2 scenarios

eline Intermediate Optimistic Mixed

.070 188.093 194.800 194.79224 92.402 99.330 99.33046 95.692 95.474 95.4621 7.358 7.679 7.7349 1.087 1.094 1.08882 45.566 46.237 46.566.514 167.397 173.969 174.032

21.6 22.4 22.4

60 C.J. Shepherd et al. / Aquaculture 467 (2017) 49–62

is not an issue. The European salmon industry imports non-GM soyamainly from Brazil as already crushed concentrate (see Section 3.2.1),paying a premium which may increase over time. Hence the strategiclogic of avoiding over-reliance on imported soya products by greaterfocus on locally grown protein substitutes, including legumes, beansand peas, aswell as focusing on novel sources of EPA andDHA, especial-ly fromalgal fermentation and transgenic oilseed crops. Ifmicrobial pro-tein (e.g. FeedKind™) is shown to be cost-effective and available on alarge scale, this could potentially start to replace fishmeal and SPC insalmon feeds from as early as 2018.

8. Overall conclusions

8.1 The Scottish salmon farming industry occupies a distinctive in-ternational market position supplying a differentiated productrange to offset potentially higher production costs comparedwith the much larger and more standardised Norwegian indus-try. Scottish production therefore relies on a variety of feed prod-ucts, including specialist formulations with a higher thanstandard content of marine ingredients and omega-3 LC-PUFA,etc. Despite its recognised economic and social benefits, thishas raised questions about the industry's environmental sustain-ability in regard to salmon feed.

8.2 Finite marine feed resources are now being replaced by plant-based ingredients in salmon feed, hence enabling salmon farm-ing to achieve net fish protein production. Although it is continu-ing to fall, the relatively higher marine content of some Scottishfeeds (averaging around 40% marine and 60% plant ingredients)compared with Norwegian farmed salmon reinforces a ‘healthyand natural’ reputation and may be seen as a necessary trade-off against increased feed costs.

8.3 There is no evidence that terrestrial agricultural animal and plantfeed resources are more sustainable for farming salmon thanusing feed ingredients based on wild-caught marine resources,provided they are sourced from sustainably managed stocks (orfrom the growing proportion of fish processing by-products,which currently represent 25–30% of marine ingredients);hence an increased utilisation of plant ingredients in salmonfeed may not be more environmentally sustainable.

8.4 The Scottish salmon industry has sought to adopt best farmingpractice in response to its producer organisation and to supplychain pressures, including responsible sourcing. Scottish certifi-cation focus on sustainable raw materials has helped to counterclaims of an unsustainable use of marine ingredients, hence thegrowing recognition that sustainability issues for farmed Scottishsalmon are nowmore related to sea lice rather than feed ingredi-ents. Certified evidence on use of renewable feed ingredients isunderpinned by third party auditing. Retailers require suppliersto comply with external certification schemes as well as theirin-house standards, with the result that farming standards aredemanding and continue to be driven up. The recent ASC certifi-cation of two farm sites in Scotland suggests that this may be-come a future benchmark of responsible practice. In additionthe Scottish ethical commitment to fish welfare (via FreedomFood/RSPCA Assured status) demonstrates the industry's will-ingness to step beyond traditional boundaries of best practice.

8.5 Increased demand for fish oil from aquaculture and the humannutrition industry has coincided with acute scarcity due to ElNiño conditions affecting the anchovy fishery in Peru and Chile.In the short-term this issue will temporarily resolve itself inPeru as the anchovy fishery recovers. In the medium-term inter-im solutions are needed tomanage this situation cost-effectively

and DHA from algal fermentation is now available albeit at rela-tively high cost and being studied as one solution. In the longerterm, in view of the food security and consumer health implica-tions, it is appropriate for the Scottish industry to considerreviewing its policy on GM feed ingredients for when new trans-genic plant sources of EPA and DHA become available. However,it is recognised that, use of GM feed ingredients at the presenttime would risk lost sales to France, which is the second mostvaluable export market for Scottish salmon.

8.6 There is evidence some consumers take into account the healthbenefits when choosing to buy salmon and some UK retailpacks indicate that eating a product portion will allow the con-sumer to meet international recommendations for weekly con-sumption of omega-3 LC-PUFA. This creates a risk for Scottishsalmon since the content of EPA and DHA has fallen because ofincreasing replacement of fish oil by plant oils in salmon feed.In current scarcity conditions fish oil supplementation of feedto meet the label claim is costly and difficult, but may be neces-sary to maintain industry and brand reputations. At the sametime the health attributes of Scottish salmon are not solelybased on EPA and DHA, but could take more account of the con-tent of amino acids, vitamins and minerals, etc., as well as ex-tremely low levels of environmental contaminants due toreplacing marine by plant ingredients.

8.7 Despite its potential benefits (e.g. cost savings and formulationflexibility), and its widespread use in salmon farming in theAmericas, there is strong supply chain resistance to incorporatingterrestrial by-products into salmon feeds in the UK. EU approvalhas been granted for use of PAP in aquaculture feeds, although incommercial practice thiswouldmean poultrymaterial as porcineby-products are problematic for religious and cultural reasons.However, UK retailers are currently unwilling to accept thehigh risk of a negative customer reaction. It is recognised thatthis policy restricts sourcing and runs counter to sustainabilityoptions, but is unlikely to change in the near future. The potentialfor insect-based ingredients has yet to be evaluated and devel-oped, but could meet similar consumer objections.

8.8 As regards the use of GM feed ingredients, a similar embargo ex-ists today in European salmon farming. This is despite their use inUK chicken feeds, signs of a weakening in UK consumer hostilityto the GM issue, and a more flexible approach by the EU and theEnglish and Welsh governments. Given the availability of non-GM soya at a relatively small price premium today, there is nocurrent incentive to alter this policy on grounds of protein avail-ability, although this situation could change in the future. The po-sition with omega-3 LC-PUFA is different as the likelycommercial availability of GM oilseed crop sources by around2020 from outside Europe offers a possible solution to a chronicshortage. Avoiding GM could mean that the Scottish industry isaccused of supplying a less healthy product than in those coun-tries using GM feed ingredients. It does not appear to be logisti-cally feasible for the UK fish feed and fish farming industries tomaintain both GM fed and non-GM fed supply lines at the sametime.

8.9 Although forward supply of proteins is not seen as a major issue,there should be continuing focus on cost-effective local alterna-tives to non-GM SPC in particular. Provided it becomes availableon schedule and remains cost-effective, the development of

61C.J. Shepherd et al. / Aquaculture 467 (2017) 49–62

microbial protein is of strategic interest, offering a novel alterna-tive to fishmeal and SPC of strategic interest.

8.10 As regards research bottlenecks, continuing focus is required onhow best to maintain EPA and DHA levels in farmed salmongiven the increased price and reduced availability of fish oil of ap-propriate certification status and lack of contaminants. The mostfeasible alternative source may be GM oil by 2020, but algal andmicrobial sources should be kept under review and interim solu-tions are required to manage this situation cost-effectively. Al-though alternative plant protein feed ingredients (e.g. beansand peas) are being studied, especially those that can be grownin the UK, it is important to be able to concentrate the proteinin order for it to replace SPC. The technical process and associatedlogistics therefore need to be defined and offer a cost-effectivesolution.

Acknowledgements

This paper is based on a recent study by the authors conducted on‘The production of high quality healthy farmed salmon (Salmo salar)from a changing raw material base, with special reference to a sustain-able Scottish industry’ (Shepherd et al., 2015). We acknowledge thesupport of the Scottish Aquaculture Research Forum (SARF), whichcommissioned this study (Project SP007), and a full list of all individualsand stakeholder organisations to whom we are grateful is included inthe original study report (Shepherd et al., 2015).

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.aquaculture.2016.08.021.

References

Asche, F., Bjørndal, T., 2011. The Economics of Salmon Aquaculture. Second Edition. Black-well Publishing Ltd., Hoboken (248 pp.).

Bendiksen, E.Å., Johnsen, C.A., Olsen, H.J., Jobling, M., 2011. Sustainable aquafeeds: prog-ress towards reduced reliance upon marine ingredients in diets for farmed Atlanticsalmon (Salmo salar L.). Aquaculture 314, 132–139.

Berge, G.M., Hatlen, B., Odom, J.M., Ruyter, B., 2013. Physical treatment of high EPAYarrowia lipolytica biomass increases the availability of n-3 highly unsaturated fattyacids when fed to Atlantic salmon. Aquac. Nutr. 19, 110–121.

Betancor, M.B., Sprague,M., Sayanova, O., Usher, S., Campbell, P.J., Napier, J.A., Tocher, D.R.,2015b. Evaluation of a high-EPA oil from transgenic Camelina sativa in feeds for At-lantic salmon (Salmo salar L.): effects on tissue fatty acid composition, histologyand gene expression. Aquaculture 444, 1–12.

Betancor, M.B., Sprague,M., Usher, S., Sayanova, O., Campbell, P.J., Napier, J.A., Tocher, D.R.,2015a. An nutritionally-enhanced oil from transgenic Camelina sativa effectively re-placedmarine fish oil as a source of eicosapentaenoic acid for farmed Atlantic salmon(Salmo salar). Sci. Rep. 5, 8104.

Callaway, R., Shinn, A.P., Grenfell, S.E., Bron, J.E., Burnell, G., Cook, E.J., Crumlish, M.,Culloty, S., Davidson, K., Ellis, R.P., Flynn, K.J., Fox, C., Green, D.M., Hays, G., Hughes,A., Johnston, E., Lupatsch, I., Malham, S., Mendzil, A.F., Nickell, T., Pickerell, T.,Rowley, A.F., A.F., Stanley, M., Tocher, D.R., Turnbull, J.F., Webb, G., Wootton, E.,Shields, R., 2012. Review of climate change impacts on marine aquaculture in theUK and Ireland. Aquat. Conserv. Mar. Freshwat. Ecosyst. 22, 389–421.

Cert ID Europe Limited, 2015. http://www.cert-id.eu (Accessed June 2015).CoGP, 2015. http://www.thecodeofgoodpractice.co.uk (Accessed June 2015).Crampton, V.O., Nanton, D.A., Ruohonen, K., Skjervold, P.-O., El-Mowafi, A., 2010. Demon-

stration of salmon farming as a net producer of fish protein and oil. Aquac. Nutr. 16,437–446.

CSIRO, 2013. http://www.csiropedia.csiro.au/display/CSIROpedia/DHA+canola (AccessedJune 2015).

FAO, 2014. The State of World Fisheries and Aquaculture 2014. Food and Agriculture Or-ganization of the United Nations, Rome (223 pp.).

FAO, 2015. http://www.fao.org/fishery/code/publications/guidelines/en (Accessed June2015).

Gail Smith, B., 2008. Developing sustainable food supply chains. Philos. Trans. R. Soc. Lond.Ser. B Biol. Sci. 363, 849–861.

Galli, C., Risé, P., 2009. Fish consumption, omega 3 fatty acids and cardiovascular disease.The science and the clinical trials. Nutr. Health 20, 11–20.

Gatlin III, D.M., Barrows, F.T., Brown, P., Dabrowski, K., Gibson, G.T., Hardy, R.W., Elliot, H.,Hu, G., Krogdahl, A., Nelson, R., Overturf, K., Rust, M., Sealey, W., Skonberg, D., Souza,E.J., Stone, D., Wilson, R., Wurtele, E., 2007. Expanding the utilization of sustainableplant products in aquafeeds: a review. Aquac. Res. 38, 551–579.

GOED, 2014. Global Recommendations for EPA and DHA Intake (Rev 16 April 2014). pp.20 www.goedomega3.com/index.php/files/download/304 (Accessed June 2015).

Harvest, M., 2014. Salmon Farming Industry Handbook 2014. Oslo (84 pp.).Hatlen, B., Berge, G.M., Odom, J.M., Mundheim, H., Ruyter, B., 2012. Growth performance,

feed utilisation and fatty acid deposition in Atlantic salmon, Salmo salar L., fed gradedlevels of high-lipid/high EPA Yarrowia lipolytica biomass. Aquaculture 364–365,39–47.

Henriques, J., Dick, J.R., Tocher, D.R., Bell, J.G., 2014. Nutritional quality of salmon productsavailable from major retailers in the UK: content and composition of n-3 long-chainpolyunsaturated fatty acids. Br. J. Nutr. 112, 964–975.

Hixson, S.M., Parrish, C.C., Anderson, D.M., 2014. Full substitution of fish oil with camelina(Camelina sativa) oil, with partial substitution of fish meal with camelina meals, indiets for farmed Atlantic salmon (Salmo salar) and its effect on tissue lipids and sen-sory quality. Food Chem. 157, 51–61.

HM Government, 2014. https://www.gov.uk/government/news/new-national-diet-and-nutrition-survey-shows-uk-population-is-eating-too-much-sugar-saturated-fat-and-salt (Accessed June 2015).

Hognes, E.S., Ziegler, F., Sund, V., 2011. Carbon footprint and area use of farmed Norwegiansalmon. SINTEF Report No F21039. Trondheim, SINTEF Fisheries & Aquaculture (30 pp.).

IFFO, 2014. The marine ingredients organisation: fishmeal and fish oil statistical yearbook2014. IFFO [online] www.iffo.net (Accessed June 2015).

IFFO, 2015. http://www.iffo.net/iffo-rs (Accessed June 2015).IUCN, 2015. http://www.iucn.org/about/work/programmes/species/our_work/the_iucn_

red_list/ (Accessed June 2015).Jackson, A.J., Shepherd, C.J., 2012. The future of fishmeal and fish oil. In: Ryder, J.,

Ababouch, L., Balaban, M. (Eds.), Second International Congress on Seafood Technol-ogy on Sustainable, Innovative and Healthy Seafood FAO Fisheries and AquacultureProceedings. No. 22. Food and Agriculture Organisation, Rome, pp. 189–208.

Kitessa, S.M., Abeywardena, M., Wijesundera, C., Nichols, P.D., 2014. DHA-containing oil-seed: a timely solution for the sustainability issues surrounding fish oil sources of thehealth-benefitting long-chain omega-3 oils. Nutrients 6, 2035–2058.

Kontali Analyse, 2013. Kristiansund NO-6517, Norway. New Analysis; Salmon World.http://www.kontali.com/?div_id=156&pag_id=244&art_id=1109 (Accessed June2015).

Makkar, H.P.S., Tran, G., Heuze, V., Amkers, P., 2014. State of the art on use of insects asanimal feed. Anim. Feed Sci. Technol. 197, 1–33.

Meinich, C., 2014. Fish Oil Presentation to the Market Forum (unpublished) Annual Con-ference of International Fishmeal and Fish Oil Organisation (IFFO), Vancouver, Cana-da (October 2014).

Munro, L.A., Wallace, I.S., 2015. Scottish fish farm production survey 2014. Marine Scot-land Science (53 pp.).

Naylor, R.L., Goldburg, R.J., Primavera, J.H., Kautsky, N., Beveridge, M.C.M., Clay, J., Folke, C.,Lubchenco, J., Mooney, H., Troell, M., 2000. Effect of aquaculture on world fish sup-plies. Nature 405, 1017–1024.

NRC, 2011. Nutrient Requirements of Fish and Shrimp. The National Academies Press,Washington DC (376 pp.).

OECD-FAO, 2013. Organisation for Economic Co-operation and Development/Food andAgriculture Organization of the United Nations Agricultural Outlook 2013. OECD Pub-lishing, Paris (324 pp.).

Olsen, R.L., Hasan, M.R., 2012. A limited supply of fishmeal: impact on future increases inglobal aquaculture production. Trends Food Sci. Technol. 27, 120–128.

Onozaka, Y., Hansen, H., Tveterås, R., 2012. Salmon's position among consumers. GlobalAquaculture Advocate 15, 68–70.

ProTerra, 2015. http://www.proterrafoundation.org/index.php/certification (AccessedJune 2015).

Reuters, 2015. http://www.reuters.com/article/2015/03/06/peru-fishing-idUSL1N0W82EM20150306#vVis42XLATAZr3Ch.99 (Accessed April 2015).

RTRS, 2011. http://www.responsiblesoy.org/en/ (Accessed June 2015).Ruiz-Lopez, N., Haslam, R.P., Napier, J.A., Sayanova, O., 2014. Successful high-level accu-

mulation of fish oil omega-3 long-chain polyunsaturated fatty acids in a transgenicoilseed crop. Plant J. 77, 198–208.

Ruyter, B., Rosjo, C., Einen, O., Thomassen, M.S., 2000a. Essential fatty acids in Atlanticsalmon: time course of changes in fatty acid composition of liver, blood and carcassinduced by a diet deficient in n-3 and n-6 fatty acids. Aquac. Nutr. 6, 109–118.

Ruyter, B., Rosjo, C., Einen, O., Thomassen, M.S., 2000b. Essential fatty acids in Atlanticsalmon: effects of increasing dietary doses of n-3 and n-6 fatty acids on growth, sur-vival and fatty acid composition of liver, blood and carcass. Aquac. Nutr. 6, 119–127.

Sanden, M., Stubhaug, I., Berntssen, M.H.G., Lie, Ø., Torstensen, B.E., 2011. Atlantic Salmon(Salmo salar L.) as a net producer of long-chain marine ω-3 fatty acids. J. Agric. Fd.Chem. 59, 12697–12706.

Sargent, J.R., Tacon, A.G.J., 1999. Development of farmed fish: a nutritionally necessary al-ternative to meat. Proc. Nutr. Soc. 58, 377–383.

Shepherd, C.J., Bachis, E., 2014. Changing supply and demand for fish oil. Aquac. Econ.Manag. 18, 395–416.

Shepherd, C.J., Jackson, A.J., 2013. Global fishmeal and fish-oil supply: inputs, outputs andmarkets. J. Fish Biol. 83, 1046–1066.

Shepherd, C.J., Little, D.C., 2014. Aquaculture: are the criticisms justified? II – aquaculture'senvironmental impact and use of resources, with special reference to farming Atlan-tic salmon. World Agricult. 4, 37–52.

Shepherd, C.J., Monroig, O., Tocher, D.R., 2015. Production of high quality, healthy farmedsalmon from a changing raw material base, with special reference to a sustainableScottish industry. Scottish Aquaculture Research Forum (159 pp.).

62 C.J. Shepherd et al. / Aquaculture 467 (2017) 49–62

SSPO, 2013. Sustainable Scottish Salmon. Scottish Salmon Farming Annual Report 2013.Scottish Salmon Producers' Organisation, Perth (20 pp.).

Statista, 2015. http://www.statista.com/statistics/263933/production-of-vegetable-oils-worldwide-since-2000/ (Accessed June 2015).

Sustainable Fisheries Partnership (SFP), 2015. Reduction Fisheries: SFP Fisheries Sustain-ability Overview 2015https://www.sustainablefish.org/news/articles/2015/07/25/sfp-publishes-annual-sustainability-overview-of-reduction-fisheries (Accessed June2015).

Tacon, A.G., Metian, M., 2008. Global overview on the use of fishmeal and fish oil in indus-trially compounded aquafeeds: trends and future prospects. Aquaculture 285,146–158.

Tocher, D.R., 2009. Issues surrounding fish as a source of omega-3 long-chain polyunsat-urated fatty acids. Lipid Technol. 21, 13–16.

Tocher, D.R., 2015. Omega-3 long-chain polyunsaturated fatty acids and aquaculture inperspective. Aquaculture 449, 94–107.

Torrissen, O., Olsen, R.E., Toresen, R., Hemre, G.I., Tacon, A.G.J., Asche, F., Hardy, R.W., Lall,S., 2011. Atlantic salmon (Salmo salar): the ‘super-chicken’ of the sea? Rev. Fish. Sci.19, 257–278.

Torstensen, B.E., Espe,M., Sanden, M., Stubhaug, I., Waagbø, T., Hemre, G.-I., Fontanillas, R.,Nordgarden, U., Hevrøy, E.M., Olsvik, P., Berntssen, M.H.G., 2008. Novel production ofAtlantic salmon (Salmo salar) protein based on combined replacement of fish mealand fish oil with plant meal and vegetable oil blends. Aquaculture 285, 193–200.

Turchini, G.M., Ng, W.-K., Tocher, D.R. (Eds.), 2011. Fish Oil Replacement and AlternativeLipid Sources in Aquaculture Feeds. Taylor & Francis, CRC Press, Boca Raton, p. 533.

USDA, 2015. http://www.usda.gov/oce/commodity/wasde/latest.pdf (Accessed October2015).

VKM, 2014. Benefit-Risk Assessment of Fish and Fish Products in the Norwegian Diet – AnUpdate. Scientific Opinion of the Scientific Steering Committee. VKM Report 15, Oslo(293 pp.).

Woodgate, S., 2014. Land Animal PAP for Aquaculture Feeds - Briefing Note, November2014. Foodchain and Biomass Renewables Association (FABRA) (unpublished).

Xue, Z., Sharpe, P.L., Hong, S.P., Yadav, N.S., Xie, D., Short, D.R., Damude, H.G., Rupert, R.A.,Seip, J.E., Wang, J., Pollak, D.W., Bostick, M.W., Bosak, M.D., Macool, D.J., Hollerbach,D.H., Zhang, H., Arcilla, D.M., Bledsoe, S.A., Croker, K., McCord, E.F., Tyreus, B.D.,Jackson, E.N., Zhu, Q., 2013. Production of omega-3 eicosapentaenoic acid by meta-bolic engineering of Yarrowia lipolytica. Nat. Biotechnol. 31, 734–740.

Ytrestøyl, T., Aas, T.S., Åasgård, T., 2014. Resource Utilisation of Norwegian Salmon Farm-ing in 2012–2013. NOFIMA Report No. 36/2014. The Norwegian Institute of Food,Fisheries, and Aquaculture Research, Tromsø (35 pp.).

Ytrestøyl, T., Aas, T.S., Åasgård, T., 2015. Utilisation of feed resources in production of At-lantic salmon (Salmo salar) in Norway. Aquaculture 448, 365–374.

Ytrestøyl, T., Aas, T.S., Berge, G.M., Hatlen, B., Sorensen, M., Ruyter, B., Thomassen, M.,Skontorp Hognes, E., Ziegler, F., Sund, V., Åsgård, T., 2011. Resource Utilisation andEco-efficiency of Norwegian Salmon Farming in 2010. NOFIMA Report No. 53/2011.The Norwegian Institute of Food, Fisheries, and Aquaculture Research, Tromsø (65pp.).